Sam Altman's OpenAI might help the Pentagon in developing AI weapons in future: ‘Never say never’

OpenAI Chief Executive Officer Sam Altman said he would not rule out helping the Pentagon develop a new weapons platform, the latest sign of a shift in how artificial intelligence companies think about working with the defense sector.

“I will never say never, because the world could get really weird,” Altman said on Thursday during remarks at the Vanderbilt Summit on Modern Conflict and Emerging Threats. However, Altman cautioned that he didn’t expect to work on such a platform for the Pentagon “in the foreseeable future,” unless it would be a trade-off among really bad options.

“I don’t think most of the world wants AI making weapons decisions,” he said in the conversation with Paul Nakasone, former head of the National Security Agency and a current OpenAI board member.

Defense contracts have historically been controversial with employees at consumer tech companies, including sparking significant protests inside Google in 2018. But the AI industry has recently shown more openness to such deals. OpenAI, in particular, has revised its policy on working in national security over the course of the past year, including announcing a strategic tie-up with defense-tech company Anduril Industries Inc. in December to work on anti-drone technology.

Beyond defense, Altman said the government needs to start putting more effort into learning how to use and integrate AI tools.

“I don’t think AI adoption in the government has been as robust as possible,” he said. There will be “exceptionally smart” AI systems by the end of next year, he added.

Altman and Nakasone, a retired four-star army general, were speaking ahead of the release of OpenAI’s forthcoming 03 reasoning model, which is set for next week. The audience included hundreds of intelligence and military personnel as well as academics.